all_results.json

{
    "epoch": 1.98,
    "eval_logits/chosen": -0.009708462283015251,
    "eval_logits/rejected": 0.08755116909742355,
    "eval_logps/chosen": -306.4169006347656,
    "eval_logps/rejected": -278.6773681640625,
    "eval_loss": 0.010828138329088688,
    "eval_rewards/accuracies": 0.49149999022483826,
    "eval_rewards/chosen": -0.00047626858577132225,
    "eval_rewards/margins": -0.00013606167340185493,
    "eval_rewards/rejected": -0.00034020686871372163,
    "eval_runtime": 840.6964,
    "eval_samples": 2000,
    "eval_samples_per_second": 2.379,
    "eval_steps_per_second": 0.595,
    "train_loss": 0.0102488252845022,
    "train_runtime": 3093.4739,
    "train_samples": 61135,
    "train_samples_per_second": 0.647,
    "train_steps_per_second": 0.04
}